Speed set the Voxel size for the time of exploration as large as possible. Keep in mind that between 10 cm and 0.1 cm, the difference in amount of voxels is around 100^3 or 1,000,000. This is super simplified, but keep the values for the size large (Low amount!) and go only smaller (Higher!) when needed.
Perhaps download Cinebench and google for results to see how your setup works comparatively. An activity Monitor or Task Manager allows for some exploration if the system reaches a limit.
Switch off all apps and functions that eat into your performance. Google is typically one app that asks for a lot of performance. Enabling redshift only when needed is a current circulating tip.
Please remember that tutorials can be recorded and sped up, then the commentary is placed on top. The tutorial should be slower, as the screen recording is active at the same time.
I do not want to go deeper here, as this forum is not about hardware. But perhaps I have to suggest at one point to have an Ask The Trainer show about performance optimizations of hardware stuff.
However, as a friend told me three decades ago when he got his first NeXT system: Whenever something gets faster, we put more options in our software and have it a little bit slowed down in return. I feel that is not so much different these days. Of course, the primary trend is that everything gets faster (Silicon and code), and expectations might run hot at one point.
Screen Shot 2023-05-05 at 7.46.52 PM.png
I quote here only. It is not my opinion, nor do I spoil here any secrets; it is since a while in the manual; Obviously, it is not read.
NPR is not a defined look. The term is based on excluding something that is also not defined. Or why do we, as Cinematographers, argue about Film vs. Digital?
I know it might be splitting hairs, but asking for NPR is so general that I can't provide an answer here. Redshift is, going public announcements for quite a while, creating a new NPR system. If it were just a few nodes tossed into a Node editor, it would be released by now, IMHO.
Here are two examples to get information from the Cinema 4D default Node system.
I hesitated with the UV suggestion, as that is not just a general do-this-do-that suggestion. In other words, if you have a UV solution, like UV Packaging, those can't be just rotated (well, sometimes, sometimes not). Hence why I always ask for a project file.
My suggestion to help that a little bit, keep asking questions. Besides that, set up small scenes and render privies in Stamp size. The primary stop in learning is a slow response from your "machine"; no matter the speed of your computer, it is always slow when it holds things that don't matter for that moment. I understand when little time is given, one like to have at least one thing rendered out as a reward.
A good mix is the best approach. Just a plane object and a sphere plus material and light, one light! Never toss too many things in a scene. A typical mistake is the illusion of progress by loading things in the scene because one has them. (… based on my observation over nearly two decades of mentoring artists)
Keep it super simple. Then explore while rendering in 1/4 of HD or smaller, just enough to get feedback. The more you learn, the faster you will be. No beefy CPU will replace this. 😉 (I have rendered as preview three projects this evening in 480x270 to get an idea of the speed and progress of a scene in terms of color and animation. That makes me faster than anyone who always renders as large as possible. I learned that in 2004 from an ILM artist.
The more you get familiar with the mix of parameters, the faster you get where you like to be, and so, the little time you have is better used. Yes, that is a little bit slower at first.
Your question, some materials can be reproduced quickly, but some need more complex calculations. This is all I can say, based on a general question.